The literature on teenage driving indicates that teenagers are at elevated risk for motor vehicle crashes. In 2003, teenagers (ages 13-19) accounted for 10% of the U.S. population, and 13% of motor vehicle crash deaths. In 2001, motor vehicle crashes were the top reason for death among the teenagers. 41 % of the female teenage fatalities and 34% of the male teenage fatalities resulted from a motor vehicle crash. Various studies have attributed this increased risk level to factors such as the fact that teenagers have low level of driving experience, tend to drive at riskier times (night-time), and usually carry other teenage passengers in their car, leading to a distracting environment for the driver. Starting in 1996, states have responded to these concerns by adopting graduated driver licensing (GDL) policies to reduce traffic hazards for teenagers. Previous literature evaluating the impact of graduated driver licensing (GDL) policies of the 1990s has demonstrated the policies' effectiveness on reducing teenage involved fatal crashes. However, much remains unanswered: Why do they work? How do they work? Will the initial effects erode? How might they be improved? The difficulty in answering questions above have been due to the unavailability of good estimates of how many teenage drivers are on the roads at any given time, and how that varies with the GDL policies. In particular, there is limited evidence suggesting whether these policies achieve favorable results by improving teenage driving behavior, or by simply limiting the amount of teenage driving temporarily. To fill this gap, we will first build a model that relates the amount of teenage driving and riskiness of teenage driving to the state level GDL policies for each state-year pair using a novel econometric framework. We will then estimate the parameters of this model using data from the Fatal Accident Reporting System (PARS) and the National Highway Traffic Safety Administration's (NHTSA) General Estimates System (GES), 1990- 2005. Our primary aim is to assess the relative effectiveness of the GDL policies on limiting the amount of teenage driving and on improving teenage driving. We also aim to understand which components of the GDL policies (adult supervision, nighttime restrictions, teenage passenger limits) contribute to better driving as opposed to limited driving. Finally, we would like to study the dynamics of the GDL policies, and understand whether they have sustained long-term effects in terms of improved teenage driving behavior. [unreadable] [unreadable] [unreadable]